Near-Optimal Sample Compression for Nearest Neighbors

Abstract

We present the first sample compression algorithm for nearest neighbors with non-trivial performance guarantees. We complement these guarantees by demonstrating almost matching hardness lower bounds, which show that our bound is nearly optimal. Our result yields new insight into margin-based nearest neighbor classification in metric spaces and allows us to significantly sharpen and simplify existing bounds. Some encouraging empirical results are also presented.

Cite

Text

Gottlieb et al. "Near-Optimal Sample Compression for Nearest Neighbors." Neural Information Processing Systems, 2014.

Markdown

[Gottlieb et al. "Near-Optimal Sample Compression for Nearest Neighbors." Neural Information Processing Systems, 2014.](https://mlanthology.org/neurips/2014/gottlieb2014neurips-nearoptimal/)

BibTeX

@inproceedings{gottlieb2014neurips-nearoptimal,
  title     = {{Near-Optimal Sample Compression for Nearest Neighbors}},
  author    = {Gottlieb, Lee-Ad and Kontorovich, Aryeh and Nisnevitch, Pinhas},
  booktitle = {Neural Information Processing Systems},
  year      = {2014},
  pages     = {370-378},
  url       = {https://mlanthology.org/neurips/2014/gottlieb2014neurips-nearoptimal/}
}